A L C). Each set of reviews contained 10 reviews. Attribute information contained across the 10 reviews determines the amount of information on an attribute (high if the attribute is discussed in all 10 reviews; low if it is discussed in only two reviews), and the amount of attribute-information conflict (high if half the reviews that discussed the attribute were positive on the attribute and the other half negative; low if all reviews that discussed the attribute discussed it either consistently positively or consistently negatively). Given this, before creating the reviews, we first had to randomly determine what attributes will be discussed in each review, and the valence and extremity of each attribute discussed in each review. Consider review set 2 in Table 1 as an example: all the 10 reviews in review set 2 will discuss the attribute "Attr1" with the same valence (i.e., H A L C on "Attr1"). We first flipped a coin to decide the valence of "Attr1" in review set 2 (e.g., head is positive and tail is negative). Once the valence was determined, we flipped a coin 10 times to decide the extremity of "Attr1" in each of the 10 reviews (e.g., head is extremely positive or negative and tail is positive or negative). The attribute "Attr2" in review set 2 (L A H C) will be discussed in only two reviews (i.e., low amount of attribute information) with a different valence (i.e., high conflict of attribute information). We first determined which two reviews would discuss "Attr2" by randomly sampling two whole numbers from 1 to 10 without replacement (e.g., if the numbers 2 and 5 are sampled, then only reviews 2 and 5 will discuss "Attr2"). Next we flipped a coin to determine the valence and extremity of "Attr2" in each of the two reviews. Using this randomization, we determined the placement of all the attributes in the reviews for all the review sets (i.e., what attributes are discussed in each review of that set), and the valence and extremity of each attribute discussed in each review. Based on the results of the randomization, we created a "review design table" (see Table A1) to numerically represent the placement, the valence, and the extremely of the attributes in the 40 reviews. The texts of the reviews were written according to the review design table. The numbers in the attribute columns of Table A1 represent the extremity and valence of that attribute in reviews (the valence and extremity are represented on a 1 to 5 scale with 1 being extremely negative and 5 being extremely positive). An empty cell means that the attribute is not discussed in that review.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.